the effect of team consolidation on research collaboration and...
TRANSCRIPT
1
The effect of team consolidation on research collaboration and performance of scientists.
Case study of Spanish University researchers in Geology.
M.J. Martín-Sempere, J. Rey-Rocha, B. Garzón-García
Centre for Scientific Information and Documentation (CINDOC)
Spanish Council for Scientific Research (CSIC)
Joaquín Costa, 22. 28002 Madrid. Spain
Tel.34- 915635482. Fax 34-915642644
This is an author post-print (ie final draft post-refereeing) of the paper published in
Scientometrics, 55, 3 (2002): 377-394
Abstract
We analyse to what extent research collaboration and performance of individual scientists is influenced
by the level of consolidation of the team they belong to. A case study of Spanish Senior University
researchers in Geology is performed. Methodology is based on the combination of a mail survey
carried out among a defined set of researchers, and a bibliometric study of their scientific output.
Results provide support for the hypothesis that consolidation of research teams would result in a greater
facility to establish contacts and collaborations with colleagues, that could benefit all members of the
team, fostering their participation in funded projects and favouring their potential to publish in
international mainstream journals.
2
Introduction
Research performance of scientists is influenced by a wide range of both individual and contextual
factors. Over the last few years scientific research has been increasingly carried out within groups or
teams of scientists. Bearing that in mind, the study of factors as team size and composition, team
stability and consolidation, and scientific collaboration among researchers and teams, is of great
importance to understand research patterns and performance of scientists.
In a previous work1, we studied a sample of Spanish Senior University Geologists, with the aim of
determining to what extent productivity of researchers is influenced by the level of consolidation of the
team they belong to. In that study, productivity, measured in terms of scientific publications, was
analysed, as one of the multiple dimensions that constitute the criteria of effectiveness and performance
of researchers and research units.
Earth Sciences, and particularly Geology, are scientific disciplines where teamwork and collaboration
have a special importance. This characteristic can be considered as an indicator of maturity of research
teams2. In Spain, Geology is a scientific discipline with a great tradition that counts with numerous
well-established, internationally recognised teams. International collaboration of these groups has been
favoured by the great diversity and geological complexity of Spanish territory that has arouse the
interest of many foreign scientists, mainly French, Dutch, German and British 3.
The present paper is an extension of the above mentioned by the same authors. Here we analyse,
together with research output of scientists, their patterns of scientific collaboration. It is intended to
respond some relevant questions proposed in that previous study. Does team consolidation favour
research collaboration? Does it favour participation of scientists in funded research projects? Does it
improve research performance, measured in terms of scientific output and research collaboration? Our
goal is to contribute to understand to what extent team consolidation level affects research performance
and habits of scientists.
3
Methods
Methodology is based on the combination of a survey carried out among a defined set of Spanish
researchers, and a bibliometric study of their scientific output. The sample analysed consist of 93
scientists doing research in Geology at University faculties. The basic structural unit for analysis are
individual scientists, not research teams.
Survey data were collected in some of the seven thematic blocks of the broader, specially created, 27-
point questionnaire, which surveys different aspects of respondents' research activity, including
publishing and collaboration patterns. Selected questions from the original questionnaire are shown in
Appendix. As in our previous paper1, researchers were assigned to one of the following categories: a) C
researchers, i.e. those working within Consolidated, well-established teams; b) NC researchers, i.e.
those belonging to Non-Consolidated, not well-established teams; and c) NT (No Team) researchers,
i.e. those that are not members of any research team, who either work with different teams in different
projects or usually work alone.
Bibliometric analysis was carried out starting from bibliographic information retrieved from the CD-ROM
version of a multidisciplinary domestic database (Spanish Index on Science and Technology, ICYT), a
multidisciplinary international database (Science Citation Index, SCI), and an international database
specialised in Earth Sciences (Geological Reference File, GeoRef).
Same methodological aspects as in our previous paper1 are valid here, with the following remarks.
Differences among researchers have been investigated with regard to productivity, number and rate of
national/international collaborations, and participations in national/international R&D programmes
and/or projects.
We refer to collaboration in terms of inter-departmental and inter-institutional collaboration. Firstly,
scientific collaboration has been studied through co-authorship indicators. Indicators based on counts
of co-authored papers are often used to study the collaboration between scientists. The evaluation of
collaboration through co-authorship presents numerous advantages 4,5, although the obtained indicators
4
must be managed and interpreted with caution. In this respect, it must be taken into account that
numerous collaborations do not result in a published paper signed by several institutions, thus they
cannot be detected by co-authorship based indicators. On the contrary, there are cases in which very
peripheral or indirect interactions between scientists result in publications signed by various authors
Furthermore, collaboration has been studied, through the number of collaborations maintained with
Spanish and foreign teams, (see appendix, survey's question 22), as well as through its participation in
R&D programs and/or projects (see survey's question 23). In this respect, it is necessary to consider
that the participation in national projects not always constitutes an indicator of inter-institutional
collaboration, since this type of projects do not necessarily entails the participation of researchers from
two or more different institutions. Not therefore the participation in the rest of programs and projects
specified in the survey, since all require the participation, in addition to Spanish researchers or teams,
of foreigners. Querying experts for opinion through methods like surveying or personal interviewing
provides another way of obtaining data that allows analysing the phenomenon of scientific
collaboration from a wider perspective. Collaboration can them be considered as two or more teams
working together, sharing resources and efforts, either intellectual or physical 4. It is also true that this
method displays other disadvantages, opposed to the co-authorship based, such as difficulty in
obtaining large and representative samples of researchers’ population, or the subjectivity inherent to
any study carried out through the survey of the parties concerned. In this respect, combination of
scientist's surveying and the bibliometric analysis of their co-authored papers, provides a wider
perspective of the scientific collaboration process.
Statistical analyses were carried out with SPSS 10.0 for Windows statistical package. Figures are
shown as 'average ± standard deviation (range)'. Differences were considered as significant with a
α<0.05. Non-parametric tests for sample comparisons were used (the Kruskal-Wallys H test and the
Mann-Whitney U test). When relationship between quantitative variables exists, K-Means cluster
analysis has been carried out, standardising the variables and not including the 'team consolidation
level' variable on the analysis. The 'cluster membership' variable is saved as a new qualitative variable.
The correspondent Chi-square test and, if appropriate, a correspondence analysis, were performed with
variables 'cluster membership' and 'team consolidation level'.
5
Results
Research collaboration
Data on collaboration obtained from collaborative articles (those signed by two or more authors with
different institutional affiliation, i.e. pertaining to different departments and/or institutions) are
displayed in Table 1. The degree of (national, international) collaboration is the percentage of (national,
international) collaborative articles out of the total number of articles.
In general, C authors show higher figures of collaboration than NC, and the latter higher than NT, both
in absolute as in relative values. The only exceptions are the number of national collaborative articles
and the degree of national collaboration, both in SCI journals, where NC authors show higher values
than C and NT ones.
Statistically significant differences, indicated in Table 1, refer to a) the number of co-authored articles
per author: C researchers appear to be significantly more productive than NT ones, both in SCI and in
the rest of journals, without finding differences between the output of these type of articles from
authors belonging to consolidated teams and non-consolidated teams; b) the total number of
international collaborative papers: Same differences as in previous case have been found when
analysing the whole set of journals, being differences mainly due to bilateral collaboration with teams
from other European Union (EU) member countries. In SCI journals, however, C researchers published
a significantly higher number of articles co-signed with foreign scientists, than their NC and NT
colleagues; and c) the rate of international to total co-authored papers and the degree of international
collaboration (both in the whole set of journals; articles published in SCI journals do not reveal
differences in these indicators).
6
Table 1: Collaborative articles per author in journals covered by SCI, GeoRef and ICYT
C NC NT Total All journals
- Collaborative articles ** 2.9±2.7 (0-11) 1.9±2.3 (0-7) 1.2±1.4 (0-4) 2.5±2.5 -- National collaboration 1.9±2.3 (0-11) 1.7±2.1 (0-6) 1.1±14 (0-4) 1.7±2.1 --International collaboration ** 1.2±1.6 (0-8) 0.5±0.9 (0-3) 0.1±0.3 (0-1) 0.9±1.45 -- Bilateral (with EU-teams)** 0.8±1.3(0-7) 0.4±0.7(0-2) 0.1±0.3(0-1) 0.6±1.2 -- Bilateral (with non-EU teams) 0.2±0.5(0-2) 0.1±0.3(0-1) 0 0.2±0.4 -- Multinational 0.2±0.6(0-3) 0.1±0.3(0-1) 0 0.1±0.5 - International / Total ** 0.38±0.36 0.28±0.36 0.12±0.32 0.33±0.36 - Degree collaboration 0.56±0.36 0.49±0.40 0.37±0.34 0.52±0.36 - Degree national collaboration 0.33±0.31 0.31±0.32 0.29±0.28 0.32±0.31 - Degree international collaboration * 0.23±0.30 0.17±0.30 0.08±0.27 0.19±0.30
SCI journals - Collaborative articles ** 1.3±1.4 (0-6) 1.2±2.2 (0-7) 0.4±0.7 (0-2) 1.1±1.5 -- National collaboration 0.7±1.0 (0-5) 1.2±1.8 (0-5) 0.4±0.7 (0-2) 0.7±1.2 -- International collaboration * 0.8±1.3 (0-7) 0.2±0.8 (0-3) 0.1±0.2 (0-1) 0.6±1.3 -- Bilateral (with EU-teams) 0.5±1.2(0-6) 0.1±0.5(0-2) 0.1±0.2(0-1) 0.4±1.0 -- Bilateral (with non-EU teams) 0.1±0.4(0-1) 0 0 0.1±0.3 -- Multinational 0.1±0.4(0-2) 0.1±0.3(0-1) 0 0.05±0.3 - International / Total 0.42±0.42 0.20±0.32 0.20±0.45 0.37±0.41 - Degree collaboration 0.62±0.37 0.60±0.42 0.55±0.53 0.61±0.40 - Degree national collaboration 0.33±0.35 0.48±0.40 0.44±0.53 0.37±0.38 - Degree international collaboration 0.29±0.38 0.11±0.21 0.11±0.33 0.24±0.36
Significant differences: * C>(NC=NT); ** C>NT
Table 2 shows the average number of collaborations per author within the different groups, according
to survey's data. In accordance with bibliometrics analysis, survey likewise shows that, in general, C
researchers collaborate more than the rest, both in national and international collaboration. In this
respect, stability and consolidation of teams seem to have a favourable effect over collaboration,
particularly over international collaboration. Spanish University Geologists belonging to consolidated,
well-established teams, performed a significantly higher average number of international collaborations
during the period studied, than those belonging to non-consolidated teams and those not belonging to a
team. Besides these absolute figures, the rate of international collaboration with regard to the total
(more than 50%), is also higher. Differences are mainly due to bilateral collaborations with teams from
other EU member countries. On the other hand, NT researchers show some figures slightly higher than
the corresponding to scientists in non-consolidated teams. This is particularly noticeable in the case of
domestic and multinational collaborations.
7
Table 2: Average number of collaborations of researchers surveyed (Source: survey's question 22)
Collaboration type C NC NT Total - National (with other Spanish teams) 3.0±2.6 (0-10) 2.3±2.0 (0-7) 2.9±1.6 (0-6) 2.9±2.4 - International * 3.6±3.4 (0-15) 1.4±1.4 (0-5) 1.7±2.0 (0-8) 2.9±3.1 -- Bilateral (with EU-teams) * 2.0±2.1 (0-10) 0.8±1.0 (0-4) 0.7±1.1 (0-3) 1.6±1.9 -- Bilateral (with non-EU teams) 0.8±1.3 (0-6) 0.5±0.6 (0-2) 0.5±0.9 (0-3) 0.7±1.1 -- Multinational 0.9±1.3 (0-4) 0.1±0.3 (0-1) 0.4±0.6 (0-2) 0.7±1.1 - International/ Total ** 0.53±0.30 0.37±0.29 0.35±0.29 0.47±0.30 - Total 5.6±3.5 (0-15) 3.7±3.0 (0-10) 4.6±2.8 (2-13) 5.1±3.4 Significant differences: * C>(NC=NT); ** C>NT
Participation in research projects and/or programmes
As indicated in Table 3, significant differences have been found in the total number of participations in
research projects and/or programmes, as well as in the number of international bilateral projects with
EU-based teams and in the number of participation in EU-funded projects. C scientists participated in a
significantly higher number of projects than NC ones. NT researchers occupied an intermediate
position, showing a higher number of participation in projects than NC ones. NT researchers overcome
their NC colleagues in the number of both national and international projects.
Most outstanding differences have been identified in the participation in international projects, which
constitute another indicator of international collaboration. C researchers overcome their NC and NT
colleagues, both on the amount of projects in which they participate and on the percentage of
international projects with regard to the total number of projects they participate in. This difference is
mainly focused on bilateral projects in collaboration with EU teams.
Table 3 Average number of participations in projects and/or programmes (Source: survey question 23)
Type of project/programme C NC NT Total - National 3.1±1.69 (0-8) 2.5±2.3 (0-7) 3.4±2.4 (0-8) 3.1±1.9 - International * 2.6±2.5 (0-10) 0.5±0.8 (0-2) 1.1±1.5 (0-5) 2.0±2.3
-- International bilateral (with a EU-team) *
1.0±1.3 (0-5) 0.4±0.7 (0-2) 0.5±0.9 (0-3) 0.8±1.2
-- International bilateral (with a non-EU team)
0.3±0.7 (0-3) 0 0.4±0.9 (0-3) 0.3±0.7
-- EU ** 0.8±1.4 (0-7) 0 0.3±0.4 (0-1) 0.5±1.2 -- Multinational 0.5±1.0 (0-4) 0.1±0.5 (0-2) 0.1±0.3 (0-1) 0.4±0.9
- International/Total * 0.39±0.26 0.19±0.32 0.22±0.24 0.33±0.28 - Total ** 5.6±2.9 (1-13) 3.1±2.3 (0-7) 4.5±2.8 (0-9) 5.0±2.9 Significant differences: * C>(NC=NT); ** C>NC
8
Characterising researchers by its scientific output and collaboration figures
In this chapter, results of cluster analysis carried out with the analysed variables are shown. The aim is
to determine whether is possible to create homogeneous clusters of individuals (researchers)
characterised by a similar performance of the concerned variables, and to what extent these groups
correlate to the ones defined according to the level of consolidation of research teams (C, NC, NT). For
this purpose, we select those variables where significant differences have been found, along with data
of scientific output from our previous study1.
The different international collaboration figures and patterns revealed by the survey allow us to
establish a first division between researchers belonging to consolidated research teams and the rest of
their colleagues (see Figure 1). A cluster analysis has been performed with four survey variables that
shows this collaboration, both in absolute or relative values. Significant differences have been found on
the belonging of scientists to clusters according to these variables (Chi-square=11.6 α=0.02). Results
of the correspondence analysis performed with variables 'cluster membership' and 'team consolidation
level' is shown in Figure 1. The corresponding plot shows that the horizontal dimension, i.e. that
accounting for the largest part of the association between rows and columns, seems to be largely
determined by C versus NC and NT researchers. NC and NT researchers are shown ascribed to lower
cluster score in all variables (cluster 3), opposite to authors belonging to consolidated teams, associated
to clusters 1 and 2. Differences among groups seems to be mainly produced by the number of
international collaborations and projects (the variables that differ the most across the three clusters),
rather than by the percentage they represents with respect to the total number of collaborations and
projects.
9
Cluster 1 2 3 International collaborations 0.03 2.1 -0.6 International projects 0.2 1.2 -0.1 International/total collaborations 0.6 0.6 -0.8 International/total projects 0.6 0.9 -0.8
Figure 1: Correspondence analysis row and column plot for cluster membership according to international collaborations and participation on international projects. Source: survey's questions 22 and 23. Table shows final cluster centres. Cluster 3 shows the lowest values for all variables, being the differences more relevant for the number of international collaborations (ANOVA F=106.7) and the number of participations in international projects (F=72,2), than for the corresponding relative values of participations in projects (F=56.0) and collaborations (F=37.2). In cluster 3, values of variables are among 0.5 and 0.8 standard deviation units below the mean for all researchers. Both dimensions of the correspondence analysis plot explain 45.3% of total variation, dimension 1 accounting for 94.5% of it.
Cluster analysis of the number of collaborations and the percentage of international collaborations,
measured through co-authored articles, has also been carried out. Although the correspondence
analysis performed with variables 'cluster membership' and 'team consolidation level' is on the limit of
statistical significance (α=0.051), it shows a picture (Figure 2) similar to that of the previous case
(Figure 1). The horizontal dimension seems to be likewise determined by C versus NC and NT
researchers, being the latter associated to lower values in all variables (cluster 3). Differences among
groups seem to be mainly produced, in this case, by the percentage of international collaborative
articles.
10
Cluster 1 2 3 Co-authored articles 1.8 0.02 -0.1 International/total co-authored articles -0.2 1.5 -0.6
Figure 2: Correspondence analysis row and column plot for cluster membership according to the number of co-authored articles and the percentage of international co-authored articles. Source: bibliographic databases. ANOVA: rate of international to total co-authored articles F=107.0; number of co-authored articles F=67.2. Both dimensions of the correspondence analysis plot explain 40% of total variation, dimension 1 accounting for 90.2% of it.
A similar division of researchers is obtained when trying to characterise them not only according to
their collaboration figures but also by their scientific output (Figure 3). The analysis carried out starting
from the data obtained from bibliographic databases, coincide in distinguishing C researchers form the
rest. Significant differences have been found on the belonging of scientists to clusters according to
these variables (Chi-square=10.8; α=0.03). As in previous figures, the horizontal dimension seems to
be largely determined by C versus NC and NT researchers, being the latter associated to lower values
in all variables. Differences among groups seems to be mainly produced by the number of articles in
both domestic and foreign journals, rather than by the number of collaborative papers.
11
Cluster 1 2 3 Articles in domestic journals 0.2 1.9 -0.3 Articles in foreign journals 1.6 0.6 -0.4 Domestic collaborative articles 0.6 1.6 -0.4 International collaborative articles 0.8 -0.1 -0.3
Figure 3: Correspondence analysis row and column plot for cluster membership according to scientific output in domestic and foreign journals together with national and international collaborative articles. Source: bibliographic databases. ANOVA: number of articles in foreign journals F=63.3; number of articles in domestic journals F=47.7; number of national collaborative articles F=41.5; number of international collaborative articles F=14.2. Both dimensions of the correspondence analysis plot explain 36.6% of total variation, dimension 1 accounting for 99.7% of it.
Similar result is obtained if analysis is only reduced to the variables that represent the scientific output
in foreign journals and the number of articles on international collaboration (see Figure 4).
12
Cluster 1 2 3 Articles in foreign journals 2.1 0.3 -0.4 International collaborative articles 0.1 1.4 -0.5
Figure 4: Correspondence analysis row and column plot for cluster membership according to scientific output in foreign journals and international collaboration. Source: bibliographic databases. Cluster 3 shows the lowest values for both variables, being the differences relevant for both variables (ANOVA F=98.89 and 109.64, respectively). Both dimensions of the correspondence analysis plot explain 47.7% of total variation, dimension 1 accounting for 91.8% of it.
Equivalent analysis carried out with the variables obtained throughout the survey shows a more acute
differentiation, dividing three groups (Figure 5). NT researchers appear associated to cluster with lower
values on scientific output and intermediate values on international collaboration (cluster 2). NC
researchers, with a higher value on output in foreign journals, show however the lowest values on
international collaboration. Finally, researchers in consolidated teams are shown associated to cluster 1,
with the higher values on international collaboration and intermediate values on publishing articles in
foreign journals. Significant differences have been found on the belonging of scientists to clusters
according to these variables (Chi-square=19.8, α=0.00). The significance of the correspondence
analysis is noticeable higher than in previous cases, as both dimensions of the correspondence analysis
plot explain 70.8% of total variation.
13
Cluster 1 2 3 Articles in foreign journals 0.1 -0.7 0.5 International collaborations 0.4 -0.5 -0.6
Figure 5: Correspondence analysis row and column plot for cluster membership according to scientific output in foreign journals and international collaboration. Source: survey's questions 13 and 22. ANOVA: number of articles in foreign journals F=71.7; international collaborations F=60.7. Both dimensions of the correspondence analysis plot explain 70.8% of total variation, dimension 1 accounting for 66.3% of it.
Analysis shown in figures 6 and 7 add a new variable to both previous cases, to differentiate authors
according to their scientific output in foreign journals and their international collaboration, inserting the
'participations in foreign projects' as an additional indicator of international collaboration.
In what respect absolute figures of these variables, a four-cluster analysis was necessary to find
significant differences on the belonging of scientists to clusters (Chi-square=16.1 α=0.014). Figure 6
shows correspondence analysis plot for cluster membership according to said variables. Dimension 1,
accounting for 84.4% of variability, is mainly determined by C versus NT researchers, with NC
scientists in a intermediate position, contributing almost nothing to the inertia of this dimension.
Researchers not belonging to any research team appear associated to lower cluster scores in all
variables (cluster 3). C and NC are lying along dimension 2, being NC researchers associated to
intermediate values (cluster 1) and C scientists near both clusters with highest values in all variables
(cluster 2 and 4). Differences among cluster seems to be mainly produced by the number of articles
14
published in foreign journals and the number of international collaborations, rather than by the number
of participations in international projects.
Cluster 1 2 3 4 Articles in foreign journals 5.7 6.4 1.7 16.2 International collaborations 2.7 11.4 1.3 6.5 International projects 2.0 4.6 0.9 7.2
Figure 6: Correspondence analysis row and column plot for cluster membership according to scientific output in foreign journals and international collaboration. Source: survey's questions 13, 22 and 23. ANOVA: number of articles in foreign journals F=126.9; international collaborations F=90.6; number of participations in international projects F= 26.0. Both dimensions of the correspondence analysis plot explain 56.3% of total variation, dimension 1 accounting for 84.4% of it.
With regard to relative values of these parameters (proportion of articles in foreign journals, of
international collaborations and of international projects) has not allowed characterising three clusters
that significantly correspond to the three teams defined according to consolidation level. No significant
differences have been found on the belonging of scientists to clusters according to these variables,
being the analysis on the limit of statistical significance (Chi-square=9.4; α=0.051). Nevertheless, the
correspondence analysis plot shows a picture (Figure 7) similar to that of the previous case (Figure 6).
15
The horizontal dimension, accounting for 91.5% of the variability, seems to be likewise determined by
C and NC versus NT researchers, being the last associated to lower cluster scores in all variables
(cluster 3), while NC and C ones appear associated to highest values in all variables (clusters 1 and 2).
Differences among clusters seem to be mainly produced by the percentage of participation in
international projects and the percentage of articles in foreign journals.
Cluster 1 2 3 % articles in foreign journals 0.3 1.3 -0.6 % international collaborations 0.9 0.01 -0.5 % international projects 1.0 0.3 -0.7
Figure 7: Correspondence analysis row and column plot for cluster membership according to scientific output in foreign journals and international collaboration. Source: survey's questions 13, 22 and 23. ANOVA: percentage of articles in foreign journals F=42.1; percentage of international collaborations F=32.6; percentage of participations in international projects F= 55.9. Both dimensions of the correspondence analysis plot explain 41.5% of total variation, dimension 1 accounting for 91.5% of it.
16
Discussion and conclusions
Same regards as in our previous paper1 concerning a) multidimensionality of research effectiveness and
performance, b) aspects to bear in mind when comparing with other studies' results, and c) the specific
characteristics of the population under study, should be considered here.
International collaboration seems to be intimately linked to, or favoured by, the degree of team
consolidation, so that the belonging to a consolidated, well-established team, entail an substantial
improvement of scientists' capacity of facility for establishing collaborations and participating in
international projects. On the other hand, it should be emphasised that, although differences are not
always statistically significant, survey's results suggest that researchers not belonging to teams
collaborate more than their colleagues in non-consolidated teams, both at the national and international
level, and participate in more international projects. Nevertheless, as inferred from bibliometric data,
this does not result in a higher number of co-authored papers. This could be interpreted as NT
researchers having a larger movement freedom to join research teams (consolidated or not), together
with a greater necessity of joining these teams to participate in funded research projects. The exception
seem to be bilateral collaborations with foreign teams, where a more direct and narrower contact
between two research teams might be the norm, while in multilateral collaborations the incorporation
of an isolated researcher is more feasible. Scientists in non-consolidated teams could have more
difficulties to establish international collaborations than NT ones, and of course, than researchers
belonging to consolidated, well-established teams. In some way, it could be said that NC researchers
are “own team's prisoners”, in a way that if the team is not able to get funded projects, all researchers of
the team remain without project. In this respect, not belonging to a team would suppose some kind of
advantage over the belonging to non-consolidated teams.
Considering jointly the results of our previous paper on research productivity1 and those of the present
study, the most significant finding is the greater scientific activity, measured in terms of productivity,
international collaboration and participation in international projects, of scientists that belongs to a
consolidated, well-established research team, opposite to those who do not. Most of the multivariate
statistical analyses performed allow us to establish a distinction between the former and the latter
17
Nevertheless, the joint analysis of the productivity in foreign journals, international collaborations, and
participation in international projects, indicate that a further differentiation could be established among
NC and NT researchers.
The lower performance of NT scientists can be due to the need to assume in solitaire all tasks inherent
to scientific research and collateral activities (manuscript writing, fund raising, administrative work,
relationship with colleagues and so forth). In this sense, the capacities of scientists who belong to a
team are harnessed, resulting in a better personal performance.
On the other hand, the labour situation of scientists is another factor important that affects their activity
and performance. In fact, our previous study showed that team size does not appear to be as important
as the number of scientist of the permanent staff within the team, that is to say the number of team's
components who have a permanent or stable job position. In this sense, the results show the positive
influence exerted on the activity and performance of scientists, by the labour stability and the fact of
carrying out their research activity within a team, preferably if this is a consolidated, well-established
one, where scientists have common objectives and interests, and where a more effective task sharing is
likely to result in an improvement of both individual and team performance.
In conclusion, the present results provide support for the hypothesis that consolidation of research
teams would result in a greater facility to establish contacts and collaborations with colleagues, that
could benefit all members of the team, fostering their participation in funded projects and favouring
their potential to publish in international mainstream journals.
18
Appendix
Selected questions of the survey 'Research activities in the R&D system in Earth Sciences in Spain:
1990-1994'.
Q6. Are you member of a consolidated, well-established research team?
• Yes, I belong to a consolidated, well-established team.
• No, my research team cannot be considered as a consolidated, well-established one.
• I am not member of any research team. Depending on the project, I work with different
teams.
• I usually work alone.
Q13. Please give the number of the following documents you produced (signed) during the
reference period (1990-1994)
• Journal articles (in Spanish journals).
• Journal articles (in US journals).
• Journal articles (in EU journals).
• Journal articles (in Latin-American journals).
• Journal articles (in other foreign journals).
• Scientific and technical reports.
• Conference papers (national).
• Conference papers (international).
• National patents.
• International patents.
• Maps.
• Books.
• Book chapters.
• Other (please specify).
19
Q.22. Please give the number of collaborations you maintained during the mentioned period
(1990-1994), with…
• Other Spanish teams.
• EU teams.
• Non-EU teams.
• Multinational teams (teams including researchers from more than one foreign country).
Q.23. Please give the number of participations, during the mentioned period, in ...
• National Programs/Projects.
• Bilateral Programs/Projects in collaboration with a EU country.
• Bilateral Programs/Projects in collaboration with a non-EU country.
• EU Programs/Projects.
• Multinational Programs/Projects.
20
Acknowledgements
Authors express their thanks to all the survey's respondents. We acknowledge Laura Barrios, at the
CSIC Department of Operational Research and Applied Statistics, for her valuable statistical advice.
Mercedes Planas is thanked for her help in writing the English manuscript.
21
References
1. J. Rey-Rocha, M.J. Martín-Sempere and B. Garzón. Research productivity of scientists in
consolidated vs. non-consolidated teams: the case of Spanish University Geologists.
Scientometrics (forthcoming)
2. D. Beaver and R. Rosen. Studies in scientific collaboration. Part I. The professional origins of
scientific coauthorship. Scientometrics, 1 (1978) No 1, 65-84.
3. J.A. Vera Torres. Análisis crítico sobre las Ciencias de la Tierra en España. In: Prospectiva en
Ciencias de la Tierra. Encuentro sobre prospectiva en Ciencias de la Tierra. Universidad
Internacional Menéndez Pelayo, 1986, 30 June-2 July, Santander. Madrid, Gabinete de Estudios de
la Presidencia, Consejo Superior de Investigaciones Científicas (CSIC), 1987, p. 7-15.
4. J.S. Katz and B.R. Martin. What is research collaboration?. Research Policy, 26 (1997) No. 26, 1-
18.
5. K. Subramanyam. Bibliometric studies of research collaboration: A review. Journal of Information
Science, 6 (1983), No. 6, 33-38.